Isolation of stem-like cells from spontaneous feline mammary carcinomas: Phenotypic characterization and tumorigenic potential

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

Author's personal copy

Research Article

Isolation of stem-like cells from spontaneous felinemammary carcinomas: Phenotypic characterization andtumorigenic potential

Federica Barbieria, Roberto Wurtha, Alessandra Rattob, Chiara Campanellab, Guendalina Vitob,Stefano Thellunga, Antonio Dagac, Michele Cillid, Angelo Ferrarib, Tullio Florioa,⁎aSection of Pharmacology, Dept. of Internal Medicine Di.M.I., and Center of Excellence for Biomedical Research - University of Genova,Viale Benedetto XV, 2, 16132 Genova, ItalybIstituto Zooprofilattico Sperimentale del Piemonte, Liguria e Valle D'Aosta,National Reference Center of Veterinary and Comparative Oncology (CEROVEC), Piazza Borgo Pila, 16129, Genova, ItalycLaboratory of Translational Oncology, IRCCS Azienda Ospedaliera Universitaria San Martino - IST- Istituto Nazionale Ricerca sul Cancro,L.go R. Benzi, 10, 16132 Genova ItalydAnimal Facility, IRCCS Azienda Ospedaliera Universitaria San Martino - IST- Istituto Nazionale Ricerca sul Cancro, L.go R. Benzi, 10,16132 Genova Italy

A R T I C L E I N F O R M A T I O N A B S T R A C T

Article Chronology:

Received 30 December 2011Revised version received6 February 2012Accepted 7 February 2012Available online 16 February 2012

Current carcinogenesis theory states that only a small subset of tumor cells, the cancer stem cells

or tumor initiating cells (TICs), are responsible for tumor formation and progression. Humanbreast cancer-initiating cells have been identified as CD44-expressing cells, which retain tumori-genic activity and display stem cell–like properties. Spontaneous feline mammary carcinoma(FMC) is an aggressive cancer, which shows biological similarities to the human tumorcounterpart.We report the isolation and phenotypic characterization of FMC-derived stem/progenitor cells,showing in vitro self-renewal, long-lasting proliferation and in vivo tumorigenicity. Twenty-oneFMC samples were collected, histologically classified and characterized for the expression ofKi67, EGFR, ER-α and CD44, by immunohistochemistry. By culture in stem cell permissive condi-tions, we isolated, from 13 FMCs, a CD44-positive subpopulation able to survive and proliferate in

vitro as mammospheres of different sizes and morphologies. When injected in NOD/SCID mice,

FMC stem-like cells initiate tumors, generating cell heterogeneity and recapitulating the originalhistotype. In serum-containing medium, spheroid cells showed differentiation properties asshown by morphological changes, the loss of CD44 expression and tumorigenic potential.These data show that stem-defined culture of FMC enriches for TICs and validate the use ofthese cells as a suitable model for comparative oncology studies of mammary biology and testingtherapeutic strategies aimed at eradicating TICs.

© 2012 Elsevier Inc. All rights reserved.

Keywords:

Tumor-initiating cellsMammary cancer

Comparative oncologyFeline

E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

⁎ Corresponding author: Fax: +39 010 3538806.E-mail address: [email protected] (T. Florio).

0014-4827/$ – see front matter © 2012 Elsevier Inc. All rights reserved.doi:10.1016/j.yexcr.2012.02.008

Ava i l ab l e on l i ne a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ loca te /yexc r

Author's personal copy

Introduction

Naturally-occurring tumors in domestic animals represent anopportunity for comparative oncology studies for their translationpotential to human pathology [1]. Cancer is the second most fre-quent cause of death in humans and the first one in dogs and cats[2]. Considering the incidence by site, mammary gland cancer isthe most frequent (32%) in women, the first (52%) in bitches andthe third (17%) in queens [2–4]. As compared to experimental can-cers in murine models, spontaneous tumors of pets represent bettermodels of human cancer since they share similar environmental riskfactors, develop in immuno-competent organisms, offer a large pop-ulation samples, and, finally, the shorter overall lifespan of domesticanimals, associated with a more rapid cancer progression, allows amore rapid response time than humans [1]. In addition, several bio-logical,morphological and clinical features of animal tumours can beidentified in the same cancer types in humans [5].

Almost 80% feline mammary cancers are malignant, histologi-cally classified as adenocarcinomas, associated with an aggressiveclinical course and rapid metastasization, [6] and surgery is themost widely used treatment.

In addition, the lack of oestrogen dependency in most felinemammary carcinomas (FMC) [7] suggests that they could be asuitable model for hormone-independent human breast cancer[8]. Finally, similarly to the human counterpart, FMCs show over-expression of p53, cyclin A [9], epidermal growth factor receptor1 (EGFR) and 2 (HER2) [10,11], suggesting that similar tumorigen-esis mechanisms may be active in human breast cancer and FMC.

Accumulating evidence in human cancer biology supports thehypothesis that tumors are initiated and driven by a subpopula-tion of cells responsible of tumor cell hierarchy and heterogeneity,sharing several features with somatic normal stem cells (stem-likecells), such as self-renewal and pluripotency. Thus they have beennamed “cancer stem cells” (CSCs) or, due to their tumorigenic po-tential, “tumor-initiating cells” (TICs). TIC identification and char-acterization has important diagnostic, prognostic and therapeuticimplications since, according to the CSC theory, effective anti-cancer therapy requires the complete elimination of TICs fromthe tumor mass, to block the indefinite regeneration of the cancercell population and overcome chemo- and radio-resistance [12].

Candidate TICs have been identified in both haematologicaland solid human neoplasms [12]. Breast cancer is the firsthuman solid tumor in which TICs were identified as cells expres-sing high levels of CD44 and low (or none) CD24 [13]. Sincethen, TICs were identified in a wide variety of neoplasms, includ-ing prostate [14], colon [15] and pancreas [16] carcinomas, glio-blastoma [17] melanoma [18], and, possibly, pituitary adenomas[19]. Several experimental approaches have been developed to ob-tain in vitro cultures efficiently enriched in stem/progenitor cellsfrom human mammary gland tumors [20]: CD44+/CD24-/low cellsorting [21], Hoechst-dye effluxing side population by FACS analy-sis [22] and in vitro cultures in growth factor-enriched medium(stem cell permissive) as non-adherent spherical clusters, namedmammospheres. Mammospheres were obtained growing dis-persed tumor cells in selective culture conditions, to favor mam-mary stem/progenitor proliferation of cells retaining self-renewal and tumorigenicity in mice [23,24]. Shifting TICs inserum-containing medium, they differentiate, adhere to substratein monolayer and acquire epithelial lineage marker expression

[21]. On the other hand, TIC-enriched cultures display increasedresistance to radiation [25] and chemotherapeutic drugs [26].Some established human breast cancer cell lines also contain asubpopulation of cell that share some characteristics with stemcells from tumor samples [27]. However, the CD44+/CD24−/low

phenotype is not a unique and definitive breast TIC profile [28],since CD44 negative cells from pleural effusion also generatemammospheres in vitro and tumors when transplanted in immu-nodeficient mice [29].

Mechanisms of tumor development seem to be similar in humanand companion animals, but CSC research in pets is currently inearly stage of development [30]. Canine stem/progenitor cancercells have been recently isolated in hepato- and cholangio-cellularcarcinomas [31], prostatic intraepithelial tumors [32], glioblastoma[33], lung adenocarcinoma [34] and osteosarcoma cell lines [35].Furthermore, cells with stem cell-like properties were isolatedfrom canine normal and neoplastic mammary gland [36] and mam-mary carcinoma cell lines [37]. To date, only putative TICs have beenidentified from FMC cell lines by spherogenesis assay [38] while nostudies rely on isolation of TICs from feline surgical tissues.

The availability of CSC derived from spontaneous tumors ofcats and dogs offers many advantages for the biological and phar-macological characterization of these cells; the bioptic material re-flects the natural setting of tumor cells, allows a relatively rapidfollow-up of clinical cases, and retains the individual heterogene-ity of drug responses, thus representing a reliable pre-clinicalmodel to assay drug targeting to TICs [39].

Most molecular bases of tumorigenesis and altered intracellu-lar pathways are common in humans and companion animals[40], thus, comparative oncology could be useful approach in CSCfield [41]. The isolation of canine and feline TICs and the study oftheir biology represents a good pre-clinical model of cancer,whose deficiency often hampers the translation of cancer biologyfindings into clinical oncology practice [1].

Current experimental models (cancer cell lines, transgenic andxenograft rodent tumors), despite their long-lived contribution tocancer studies, are not completely able to reproduce the high indi-vidual heterogeneity within each tumor histotype and among pa-tients [5]. Thus, pets may represent a new tool to bridge rodentexperimental models to human cancer stem cells.

In this study, exploiting human breast cancer stem cell knowl-edge, we identified and cultured putative TICs from fresh FMC tis-sues, displaying phenotype and biology features of CSCs:expression of CD44, ability to grow as mammospheres, self-renew under stem cell-permissive culture conditions, differentia-tion when shifted in serum-containing medium in vitro. The for-mation of sub-cutaneous FMC xenografts in NOD/SCID miceconfirmed the in vivo tumorigenic potential of isolated cultures.

This is the first report demonstrating the presence of TICs inFMC, supporting the relevance and the feasibility of studying petspontaneous malignancies in order to elucidate the mechanismsof tumor biology.

Materials and methods

FMC samples

Twenty-one tissue samples from FMC resections were provided bythe local network of free-lance veterinary doctors. Tumor

848 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

sampling from surgical specimens of FMC were divided into 2parts: the first was fixed 10% buffered formalin, embedded in par-affin and used for histopathologic and immunohistochemical ana-lyses, the second, aseptically excised, was immediately placed intotransporting culture medium (enriched in antibiotics and antifun-gal drug, see "Primary FMC cell culture" procedure) and processedas described below.

Fixed tissue section (4 μm) were stained by standard haema-toxylin and eosin (H&E) method for histological examinationand classified according to the WHO International HistologicalClassification of Mammary Tumors of the Dog and Cat [42].Grade was defined according to the histological grading systemof canine and feline mammary carcinoma [43]. All malignant tu-mors were scored from 0 to 3 for tubule formation, mitotic countand pleiomorphic nuclei; the score of all three components wereadded together to give the grade.

Immunohistochemistry

Feline mammary surgical samples and tumor xenografts removedat mice sacrifice were formalin-fixed and paraffin-embedded andsectioned at 5 μm thickness, and used for immunohistochemistry(IHC). Tissue sections were deparaffinized in xilene and rehy-drated through graded alcohols, then subjected to an antigen re-trieval protocol by heating in citrate buffer (Target RetrievalSolution, citrate pH 6.0, Dako, Milano, Italy). Endogenous peroxi-dase activity was blocked for 30 min in 3% hydrogen peroxide,and for non-specific binding sites in 5% bovine serum albumin(BSA, Applichem, Darmstadt, Germany). Tissues were incubatedovernight at 4 °C in the following primary antibodies: anti-CD44(rabbit monoclonal; Cell Signaling Technology, Danvers, MA,USA), anti-ER-α (mouse monoclonal, Dako), anti-EGFR (rabbitmonoclonal; Cell Signaling Technology), and anti-Ki67 (mousemonoclonal, clone MIB-1, Dako). Slides were rinsed in Tris-buffered saline (TBS) (pH 7.4, Bioptica, Milano, Italy), incubatedwith secondary antibody Real Envision Detection System Peroxi-dase/DAB+, mouse/rabbit (Dako) according to the manufacturer'sinstruction and then stained with 3-3'-diaminobenzide. Sectionswere counterstained with Mayer's haematoxylin solution(Sigma-Aldrich). Negative controls were run in parallel, substitut-ing the primary antibody with 5% BSA in TBS or rabbit/mouse IgG(Sigma-Aldrich). As positive controls, tissue sections of humanbreast cancers were used. Immunostaining was evaluated bylight microscopy (Nikon Coolscope, Firenze, Italy) to determinethe antigen expression and localization. For CD44, ER-α andEGFR expression, both the intensity of immunoreaction and thepercentage of positive cells were evaluated, and tissues were grad-ed on a scale of 0 to 3 to determine the score of immunopositivity.Ki67 expression index was evaluated as % of positive cell per totaltumor cell (at least 1,000) in 10 randomly selected microscopicfields.

All IHC scoring were carried out by blinded examination ofslides, performed independently by two observers (AR and CC)to ensure that the semi-quantitative estimate of stained percent-age of cells was as consistent as possible

Primary FMC cell cultures

Feline mammary tissues, from the veterinary operating room,were collected within 4 h from the surgical resection and

transported on ice in sterile tubes in medium DMEM/Ham's F12(1:1, Lonza, Verviers, Belgium) containing penicillin/streptomycin(200 U/ml; Lonza), amphotericin B (250 ng/ml, Sigma-Aldrich) toprevent contamination from bacteria or fungi.

A total of 21 fresh samples were collected from 2008 to 2011.Under a biohazard hood the specimen was transferred onto a

100 mm cell culture dish and washed three times with PBS toeliminate blood and debris eventually present. Tissue was dissectedusing sterile scissors, razors, and forceps to take off skin or othernon-tumor tissues and then cut in small pieces and passed throughcell strainers (70 μm, BD Biosciences,Milano, Italy) recovering singlecells by centrifugation.

Occasionally, minced tissues were digested with type I collage-nase/hyaluronidase (2 mg/ml, Sigma-Aldrich) for 30 min at 37 °Cand filtered through 70 μM mesh to discard clumps. The suspen-sion was centrifuged and washed once in PBS to obtain the FMCcell pellet.

Cells were splitted in 2 culture conditions: 1) serum-containing medium (SCM) DMEM/Ham's F12 (1:1) medium, 10%FBS (Lonza), penicillin/streptomycin (100 U/ml), and glutamine2 mM (Lonza) and 2) serum-free medium (SFM), a stem-cell per-missive DMEM/Ham's F12 (1:1) medium without FBS, supple-mented with epidermal growth factor (EGF, 20 ng/ml), and basicfibroblast growth factor (bFGF; 10 ng/ml) both from PeproTech,(London, UK), 0.4% bovine serum albumin (w/v, Sigma-Aldrich),insulin (5 μg/ml, Sigma-Aldrich). These culture conditions, enabletumor cells to retain the molecular characteristics of the primarytumor with only minor changes in differentiation, expression pat-tern, and genetic mutation profile [21,23].

To avoid fibroblast contamination and overgrowth in culture,FMC cells were purified by immunomagnetic separation withanti-fibroblast microbeads (MACS, Miltenyj Biotec, Bologna,Italy) as reported [44]. Pro-collagen (anti-procollagen type I,SP1.D8, Developmental Studies Hybridoma Bank, Iowa City, IA,USA) immunofluorescent staining was performed to verify the ab-sence of fibroblast contamination. Cells were harvested usingtrypsin-EDTA and growth factors were added to the media everyother day. Individual culture wells were photographed and scoredfor the presence of cell monolayers and mammospheres as well ascell and sphere morphologies.

Subculture of cancer stem-like mammospheres from FMC

To assess self-renewal and capacity for sphere regeneration, pri-mary cultures and mammospheres were harvested after 7 daysin culture and collected after centrifugation, dissociated by pipet-ting and re-plated in 24-well culture dishes in fresh medium atclonogenic density. The formation and the morphology of clustersand spheres as evaluated after at least 1 week of culture by micro-scopic observation and monitored for up to 21 days after initialseeding.

In vitro cell differentiation

FMC cells were examined for their differentiation ability. Spherecolonies were trypsinized and dissociated into single cells, and dif-ferentiation was induced by transfer cells, grown in stem-permissive medium SFM, in DMEM/Ham's F12 supplementedwith 10% FCS (SCM) without growth factors and culturing themfor at least 10 days.

849E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

MTT cell viability assay

The proliferative potential of FMC cells was assayed by the MTTcolorimetric assay that determines the ability of viable cells to re-duce the tetrazolium salt [3–(4,5-dimethylthizol-2-yl)-2,5-diphe-nyltetrazolium bromide] (MTT, Sigma-Aldrich) into an insolubleformazan precipitate [45]. After culture expansion, cells wereseeded in 24-well plates in SFM or SCM medium for time-courseexperiments. In differentiation assays the SFM cultured cellswere switched to SCM to determine growth rate modulation.MTT solution (2 mg/ml in PBS) was added for 4 h at 37 °C. Opticaldensity (O.D.) was measured spectrophotometrically at 570 nm.Data are expressed as mean±S.D.

Immunofluorescence

To characterize FMC cells and visualize the expression and local-ization of specific markers, immunocytofluorescence (IF) wasperformed [46]. Briefly, stem (mammospheres) and differentiat-ed cells were grown on glass coverslips, fixed in 4% paraformal-dehyde, rinsed in PBS (pH 7.4), and treated with 0.1 M glycineand Triton X-100 (0.3% in PBS) then, aspecific sites wereblocked in 10% normal goat serum (Sigma Aldrich) for 30 min.The following antibodies were applied for 1 h at r.t.: CD44(Cell Signaling Technology), epidermal growth factor receptor(EGFR, Cell Signaling Technology), ER-α (Dako), epithelialmembrane antigen/mucin1 (EMA/MUC1, Dako), cytokeratin 18(CK18, Dako). The secondary fluorescent antibodies, Alexa488–conjugated goat rabbit-specific and Alexa 568 goatmouse-specific (1:100; Molecular Probes, Invitrogen, Milano,Italy),were added for 1 h at r.t. Nuclei were counterstainedwith 4',6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) andmounted with Mowiol (Merck, Darmstadt, Germany). Negativecontrols were included in the experiments by omitting primaryantibodies or using mouse IgG (Sigma-Aldrich). IF images werecaptured using a DM2500 microscope (Leica Microsystems,Wetzlar, Germany) equipped with a DFC350FX digital camera(Leica Microsystems) or confocal laser scanning microscopy(Bio-Rad MRC 1024 ES).

In vivo tumorigenicity

To test the ability of FMC stem-like cells to initiate tumor in vivo,dissociated mammosphere cells (200,000) cultured in SFM andthe corresponding cultures in medium with 10% FBS, were resus-pended in 200 μl of matrigel (BD Biosciences) and injected intoboth flanks of 4–6 weeks old non-obese diabetic NOD⁄SCID mice(Charles River, Milano, Italy) under anaesthesia ketamine/xyla-zine cocktail. Tumor growth was monitored twice a week untiltumor appearance, then nodule diameters were measured with acaliper. Tumor volumes in mm3 were calculated by the formula:V=W2×L/2, where W is the width and L is the length of thetumor [47].

Mice were monitored for disease symptoms and were sacri-ficed by CO2 asphyxiation when they showed weight loss or anysevere sign of disease, in compliance with guidelines approvedby the Institutional Animal Care at the A.O.U. San Martino - IST,Istituto Nazionale Ricerca sul Cancro (Genova). Tumors were re-moved, fixed in formalin and paraffin-embedded before H&Estaining to determine the histological structure.

Statistical analysis

Statistical significance was assessed by ANOVA. Statistics wereperformed using SPSS 9.0 software (SPSS Inc., Chicago, USA).P≤0.05 was considered statistically significant.

Results

Histopathology and clinical data of FMC samples

The clinico-pathological features of the twenty-one primary FMCsamples analyzed are summarized in Table 1. At time of surgery,the mean age of cats was 11 years, 8 animals were intact and 13were spayed females. Histological diagnoses according to theWHO classification [42] included 19 simple carcinomas (9tubulo-papillary, 6 solid, 3 tubular, 1 papillary) and 2 cribriformcarcinomas. Grading was as follows: 4 moderately differentiated(grade 2) and 17 poorly differentiated (grade 3) tumors.

Immunohistochemistry

IHC was performed on all the tumors to evaluate Ki67 expressionindex as proliferation marker and to identify CD44-expressingcells, as potential TICs, to characterize the tumor phenotype asfar as the expression pattern of mammary carcinoma relevant re-ceptors (ER-α, EGFR).

Ki67 expression index was evaluated as percentage of positivecells (Table 2). Two FMCs showed high positivity for Ki67 antigen(>50% of cells), 5 intermediate (range 10-50% of positive cells)and 14 low positivity (>10%) (Table 2).

For the other markers, neoplastic cell immunoreactivity wasassessed using a light microscope by a semiquantitative scoreranging from not detectable (0), weak (1), moderate (2) to a

Table 1 – Clinical data and histopathology of felinemammary tumors.

Total n. 21

Mean Age 11 yrs. (range 6–16)N. %

Histology CS 6 28CT 3 14CTP 9 41CC 2 9CP 1 4

Breed European 21 100Sex F 8 38

FS 13 62Localization M1 3 14

M2 5 24M3 3 14M4 9 43NA 1 5

Grade 1 0 02 4 193 17 81

Histology: CS, Carcinoma, Solid; CT, Carcinoma, Tubular; CTP, Carcinoma,Tubulopapillary; CC, Cribriform Carcinoma; CP, Carcinoma, Papillary.Sex: F: Female; FS: Female, spayed.Localization: M1, axillary; M2, thoracic; M3, abdominal; M4, inguinal.NA: not available.

850 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

strong (3) positivity considering both cell positive percentage, in-tensity and tissue distribution of all markers. CD44 expression inFMC tissue sections was analyzed in light of previous studiesreporting the CD44+/CD24-/low phenotype associated to CSC inhuman breast carcinomas [28]. The level of positivity was evaluatedin 20 samples: 11 were negative, 8 showed a moderate immunore-activity and 1 a strong expression of CD44 (Table 2). CD44 localiza-tion was evaluated throughout the whole tissue, although thepredicted rarity of these cells and their localization in specificareas, corresponding to potential stem cell niches, may have deter-mined the occurrence of false negative tumors. In fact, in positivesamples CD44 immunostaining was detected in scattered foci oftumor cells while no immunoreactivity was found in most of thetumor mass. CD44 expression did not substantially vary among dif-ferent histological types, indicating the presence of this sub-population of cancer cells within all FMCs (Fig. 1). Conversely, theexpression of CD24 could not be assessed due to the unavailabilityof antibody cross-reacting with the feline antigen (data not shown).

Steroid hormone receptors, particularly ER-α, play a key role inthe development of mammary tumors in animals and humans.The great majority of FMC we analyzed did not show ER-α expres-sion (18/21, 86%) with only 3 cases displaying slight immunoreac-tivity (score=1) (Table 2), thus confirming the low percentage ofFMC expressing ER reported in previous studies [13]. The intracel-lular pattern of ER-α expression in positive tissues revealed amarked heterogeneity of staining, some intensely positive nucleiare close to areas with absent ER-α immunostaining (Fig. 1).

EGFR-positive breast tumors display an aggressive behaviorand high proliferation rate. EGFR immunopositivity (score=1-2)was observed in more than half of the FMC analyzed (53%).

Interestingly, several EGFR-negative tissues at the cytoplasmiclevel (score=0) showed a marked nuclear staining (8/9 cases),and 6/7 cases with membrane/cytoplasmic immunoreactivity(score=1) concomitantly express nuclear EGFR (data notshown). Representative EGFR staining is shown in Fig. 1.

Isolation and in vitro propagation of cancer stem cell-likepopulation from FMC

Then, we evaluated whether it was possible to isolate and grow invitro putative FMC stem-like cells, as previously reported inhuman breast tumors. Based on the established protocol forhuman breast cancer stem/progenitor cultures [23], we used theoptimized culturing medium containing 20 ng/mL EGF and10 ng/mL bFGF as a stem-permissive medium, favoring the growthof undifferentiated cells with self-renewal capacity.

All 21 consecutive fresh FMC specimens collected were me-chanically/enzimatically dissociated to single cell suspension andcultured both in serum-containing medium (SCM) or in serum-free medium (SFM) supplemented with EGF and bFGF. Cultureswere depleted of fibroblasts, an effective step in the establishmentof primary tumor cell cultures, to avoid fibroblast overgrowthwithin the culture and interference with both characterizationand proliferation studies. Each culture was assayed for fibroblastcontamination by immunofluorescence for pro-collagen (datanot shown) to control tumor cell purification. Purified FMC cul-tures were maintained in both SCM and SFM culture conditionsfor at least 1 week before performing further characterizations,to stabilize cultures and deplete possible residual contaminatingcells (red blood or inflammatory cells) and debris.

Thirteen out of 21 (66%) FMC samples (named FMC1-13) orig-inated primary cultures allowing the assay for the presence of astem cell subpopulation, while the remaining cultures failed togrow for different reasons, including low cell viability, bacterialcontaminations, insufficient number of collected cells.

Similarly to TICs derived from human tumors, FMC cells wereable to form spheroid colonies in serum-free, stem cell-permissivemdium. Cells grown for 48–72 h in this conditions displaied arounded morphology (Fig. 2A) while only rare cells showed anepithelial-like shape; after 96–120 h all cells round up to formanchorage-independent spheroids (mammospheres)(Fig. 2B). Fewcultures displayed small adherent colonies, readily detachablefrom tissue-culture plastic (Fig. 2C).

Corresponding primary cultures, immediately plated in serum-containing medium after the dissociation, grew as adherentmonolayers showing a predominance of epithelial-like cellsspindle in shape, (few free-floating cells but deficient in mammo-sphere formation) able to grow in culture for several passageswithout observable morphological changes (Figs. 2D-E-F ).

Repeatedly mammosphere formation reflects self-renewal andanchorage independent growth of stem-enriched breast cancercells. Seven to 14 days after plating, most cultures formedmultiplespheres although of different size and shape (some more roundedand tightly aggregated, others with a hollow morphology and ir-regular borders) (panels in Fig. 2G). FMC primary spheroids orsemi-adherent colonies were disaggregated, seeded at low-density in SFM and maintained in culture for up to 2 weeks.After disaggregation, cultures in stem medium were mainly com-posed of single floating cells, which gave rise to secondary sphereswithin about 1 week. Almost all FMC cultures were able to form

Table 2 – Immunohistochemical (IHC) analysis of FMCtissues.

Expression index* N. %

Ki67 1 14 662 5 243 2 10(ND 0 0)

IHC score° N. %

CD44 0 11 551 8 402 0 03 1 5(ND 1 -)

ER-α 0 18 861 3 142 0 03 0 0(ND 0 0)

EGFR 0 9 471 7 372 3 163 0 0(ND 2 -)

Total tumor tissues: 21.*Ki-67 expression index: 1=<10%, 2=10-50%, 3=>50% of positivecells/total tumor cells.°IHC score: 0=negative; 1=weak positivity; 2=moderate positivity;3=strong positivity.ND=not determined.

851E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

EGFR

ER-αα

CD44

H&E

100μm 50μm

Neg Neg (m) Neg (r)

Fig. 1 – Immunohistochemical expression of Ki67, CD44, ER-α and EGFR in feline mammary lesions.Representative H&E staining ofa tubulopapillary carcinoma, analyzed for the expression of CD44 (cytoplasmic staining), ER-α (distinct nuclear staining) and EGFR(both nuclear and cytoplasmic immunopositivity) as indicated by arrows.Antibody localization was done using horseradishperoxidase, with dark brown staining indicating the presence of the specific antigen. Magnification 20X (left panels) and 40X (rightpanels). Lower panels: negative controls were obtained both by omitting the primary antibody (Neg) and by using mouse (Neg m)or rabbit (Neg r) IgG as primary antibody.

852 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

secondary spheres, thus supporting the presence of viable self-renewing cells (data not shown).

Although cancer-derived spheres from single cats displayed avariable growth rate, all the samples analyzed underwent pro-longed expansion in vitro. FMC cell cultures grown in stem cell-permissive medium were serially propagated in vitro for severalpassages (up to 8–10) confirming the presence of self-renewalproperties of putative cancer stem cells. FMC proliferation wasdemonstrated by MTT assay performed at different time points(day 7, 14 and 21 from plating) and reported in Fig. 2H.

Phenotypic characterization of FMC stem-like cells

To characterize FMC cell subpopulation endowed with stem-likefeatures, 13 FMC cultures were analyzed for the expression of rel-evant markers by immunofluorescence. The number of cases ana-lyzed and the percentage of positive cases are described in theTable within Fig. 3. First, stem/progenitor cell phenotype was in-vestigated by assessing the expression of CD44, as putative dis-tinctive feature of stem-like mammary cancer cell. FMC culturesmaintained in SFM were enriched for CD44-positive cells in both

A D

EB

C F

HG

0

50

100

150

200

250

300

0 7 14 21

Gro

wth

(%

vs.

day

0)

Time (days)

FMC7

FMC8

Fig. 2 – Morphological appearance and proliferation of FMC spheres.Phase-contrast images of mammospheres obtained culturingFMC-derived cells in stem serum-free medium and corresponding primary FMC cultures grown in serum-containing medium.Panels A, B and C: formation of free-floating spheroids, mammospheres and cell clusters obtained by culturing cells isolated fromFMC in stem-permissive medium (representative cultures FMC4, FMC5 and FMC8). Original magnification 10X. Panels D, E andF: adherent monolayers grown in 10% FBS medium. Original magnification 10X. Panel G: examples of FMC spheroid morphologies(representative cultures FMC1, FMC4, FMC7, FMC8). Original magnification 20X. Panel H: representative growth curves of FMCcultures maintained in stem-permissive medium showing the in vitro proliferation potential of these cells .

853E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

spheres and partially adherent colonies in almost all cells of 12/13tumors analyzed, although some negative cells were occasionallyobserved (Fig. 3). This result demonstrates the enrichment inFMC stem-like cells (at least as far CD44 expression), in stem-cell permissive culture condition in vitro, confirming their identifi-cation in scattered areas of FMC tissues analyzed by IHC (Table 2) .

To further characterize FMC stem-like cells, cultures werestained for ER-α and EGFR (Fig. 3). We observed a marked positiv-ity for both receptors within spheres and single cells growingunder SFM conditions, in a high percentage of the cultures ana-lyzed (78% ER-α and 86% EGFR). While EGFR expression wasalso detected in a similar percentage of FMC tissue sections (86%,as reported in Table 2) more surprising was the high expressionof ER-α in isolated cells in vitro as compared to the IHC analysisof parental tumors (only 14% of positive tumors, Table 2).

Differentiation ability of FMC stem-like cells

A peculiar in vitro characteristic of CSC is the ability to differentiateupon changes in culture conditions. To determine whethermammosphere-derived cells are able to generate differentiatedcells after growth factors (EGF, bFGF) withdrawal, CD44-positivespheroids (Fig. 4A, panels i. and iv.) were disaggregated and platedin differentiating conditions using serum-containing medium. Cellcultivation in the presence of serum for about 1 week (range5–10 days) induced, in a first phase, substrate adhesion of thefloating and rounded cells, with some cell aggregates still presentin the culture (Fig. 4A, panels ii. and v.), followed by the completeadherence to the dish surface and acquisition of epithelioid mor-phology. Cell clusters were formed that progressively expandedas monolayer culture within about 2–3 weeks (Fig. 4A, panels iii.and vi.). Thus, FMC stem-like cell cultures shifted to a serum con-taining medium, acquired features resembling primary FMC cellcultures immediately after isolation, displaying epithelial-likemorphology and mostly growing as adherent monolayer(Figs. 2D, E, F).

Immunofluorescence analysis revealed that cell monolayersexpress EMA and CK18 (Fig. 4A, right panels), indicating the

presence of cells of epithelial origin, particularly the positivityfor CK18 indicates the presence of proliferating malignant epithe-lial cells.

The stem cell marker CD44, highly expressed in FMC spheres,was down-regulated in differentiated cells (Fig. 4B). In most tu-mors immunofluorescence showed increased expression of EMAin differentiated cells as compared to CSC cultures (Fig. 4B),while CK18 show very limited change in expression in this setting(data not shown). This observation corroborates the finding thatCD44 labels the most immature population in FMC spheroid cul-tures, able to expand in the presence of EGF and bFGF and to gen-erate differentiated cells under appropriate conditions.

We then compared the in vitro growth potential of stem-likeand differentiated FMC cells. While stem-enriched cultures dis-played a stable exponential growth (Fig. 4C), differentiated cells,after shifting to serum-containing medium (10% FBS) and culturedfor up to 12 days, lost the proliferative activity and they can nolonger divide as CSCs do. In fact, sustained proliferation is a pecu-liar feature of both normal and cancer stem cells. (Fig. 4C).

Tumor-initiating ability of FMC stem-like cells

To date the best identifying feature of CSCs is the operational abil-ity to phenocopy the original tumor after xenografting in mice. Toevaluate the tumorigenicity of FMC cells in vivo, cells wereinjected into NOD/SCID mice through a subcutaneous route. Mam-mosphere cells (200,000), derived from 2 tubulopapillary adeno-carcinomas (FMC12, FMC13), were dissociated andsubcutaneously injected into both flanks of NOD/SCID mice(Fig. 5A, left panel). Same number of the corresponding adherentserum-differentiated cells (Fig. 5A, right panel) were inoculatedin the same experimental conditions. FMC cells were able to en-graft in recipient mice to form palpable tumors at the site of injec-tion after 30 days (taken as reference, 100% of tumor volume). Thefollowing weeks, FMC stem-like cell xenografts grew rapidly alongthe duration of the experiments (about 1.7-fold increase after42 days) while serum-cultured cells did not (Fig. 5B). Tumor de-velopment was monitored and animals were sacrificed 54 days

FMC cultures

Positive/total (%)

CD44 12/13 (92)

ER-αα 6/9 (78)

EGFR 6/7 (86)

CD44 ER-α EGFR

EGFRER-αCD44

NEG (m) NEG (r)

Fig. 3 – Enrichment for the CD44+ cells and marker expression of FMC cells grown in stem-permissive medium.Immunofluorescence analysis of CD44, ER-α and EGFR in spheroids derived from FMC cultures. Images from fluorescencemicroscopy (upper panels, original magnification 20X) and confocal analysis (lower panels, original magnification 60X).Representative negative controls obtained by substituting the primary antibody with mouse (NEG m) or rabbit (NEG r) IgG areincluded. The table summarizes the number and the percentage of cultures positive for the relevant markers over the total. Allexperiments were performed within two to three passages.

854 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

post FMC cell graft (when the increase of the mean % volume ofserum-free derived tumors was 2.5-fold), before development ofsevere deficits for the animals due to the size of tumor masses.The large tumors detected in stem cell-injected mice invaded theperitoneum and show higher neovascularisation than those de-rived from serum-cultured cells.

A further decisive criterion for CSCs is their ability to recapitu-late the phenotype of the original tumor after mice transplanta-tion. Tumors developed in mice largely reflected the histology ofthe feline parental adenocarcinoma, i.e. malignant glandular epi-thelial lesion with a predominance of the papillary componentand extensive central necrosis (Fig. 5C). The analysis of H&E-stained paraffin-embedded sections from xenografts showed atubulopapillary histotype, with areas showing a solid compact ap-pearance, characterized by the presence of tubules and papillaryprojections with scanty stromal component and strong infiltrationinto surrounding tissues (Fig. 5C).

These results demonstrated the tumorigenic properties of FMCcultures enriched in tumor-initiating cells.

Finally, cells isolated from dissected tumor xenografts were re-cultured showing cell morphology and expression markers similarto the original culture and, under stem cell conditions and wereable to form mammospheres (data not shown) confirming thatFMC stem cells existed in xenografted tumors.

Discussion

Several experimental evidence demonstrated that solid tumors,including breast cancer, exhibit a functional cell hierarchy, com-prising differentiated cells and a stem-like subpopulation capableof initiating tumor and driving cell heterogeneity [48]. However,beside tumorigenicity assays in immunodeficient mice, phenotyp-ic and functional characterization of TICs is still under investiga-tion and several open questions are still unsolved [49,50].

The lack of appropriate investigational models for cancer cellheterogeneity, is one of the major challenges for modern experi-mental oncology. Cancer cell lines, maintained in culture

ii. iii.

iv. v.

0% FBS

timeDifferentiation time

10% FBS10% FBS monolayers

CD44

CD44

CD44

EMA

EMACD44 MO

NO

LA

YE

RS

SP

HE

RE

S

0

50

100

150

200

250

300

350

400

450

500

550

0

Gro

wth

(%

vs.

day

0)

Time (days)

Stem medium

10% FBS medium

CK18EMA

EMA CK18

A

CB

i.

vi.

12963

Fig. 4 – Cells derived from FMC spheroids grown in serum-containing medium: morphological changes, marker expression andgrowth rate potential. A. Micrographs of two representative FMC spheroids (i. and iv.) and morphological changes occurring in thepresence of serum and growth factor withdrawal, showing that cells adhered to the plastic (ii. and v.) and acquired the typicalmorphologic features of differentiated cells (iii. and vi.). Expression of EMA and CK18 antigens in sphere-derived adherentmonolayers assessed by immunofluorescence staining (magnification 20X). B. Upon the onset of differentiation FMC spheres showa decrease in CD44 marker expression while EMA staining was increased by applying medium containing serum as assessed byimmunofluorescence (red staining, magnification 20X) and confocal microscopy (green staining, magnification 60X). C. Extendedproliferative capacity of FMC-derived spheres in comparison with their differentiated counterpart (representative culture fromFMC13). Growth curve values were obtained by MTT assay at the indicated time points. While cells in stem medium displayed astable exponential growth, differentiating tumor cells did not retain the ability to proliferate and rapidly decline in number.Negative controls (as depicted in Fig. 3) were performed as described in Material & Methods (data not shown).

855E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

conditions for countless passages in the presence of serum, arehighly monomorphic, homogeneous and possibly mutated, thus,often fail to reproduce in vitro the biology of primary cancercells. In particular, many controversies were raised about celllines currently used as human breast cancer models [51] since

they do not reflect the heterogeneous phenotype of cell typeswithin breast tumors, being are poor predictors of drug efficacyin preclinical studies in vitro and in vivo. On the contrary, CSC re-search from spontaneous tumors allows a reliable model tostudy cell biology and tumor repopulation of breast cancer, retain-ing the dynamic characteristics of fresh tissue [52].

However, only small amounts of fresh tissue is obtainable fromindividual human breast carcinomas (early diagnosis luckily pre-vents the development of large tumors in most cases, and intra-and post-surgery pathological examination significantly furtherreduce the tumor mass available for experimental purposes),while technical issues concern the cell growth in vitro, since thefrequent use of neoadjuvant therapy makes the removed tumorcomposed only of cells exposed to drugs, compromising the reli-ability of the model. Thus, cell lines and rodent models are stillcentral for the study of breast cancer biology and preclinical phar-macology, notwithstanding their limitations in the faithful repro-duction of human breast tumor characteristics. On the contrary,pharmacological treatment options for FMCs are very limitedand surgical excision remains the most common therapeutic inter-vention, removing large tumors frequently involving more thanone mammary gland. The availability of untreated large samples,counterbalances the relative low incidence of the disease in cats.

Thus, naturally-occurring tumors in companion animals mightovercome issues related to the use of primary human tissues. Inaddition, FMC strikingly shares some biological and molecular fea-tures with human breast cancer [6] and FMC could be considered amodel for aggressive, locally infiltrative and metastasizinghormone-independent human breast cancer [53].

To date few studies characterized putative CSCs from compan-ion animal tumors, studying canine mammary cell lines [30,54]and tumors [36], while no data about CSCs are available in anytype of feline tumors. In this report, we addressed the intriguingquestion whether FMC stem/progenitor cells could be isolatedand cultured as reliable experimental model for both animal andhuman studies.

To this aim, we collected 21 FMC post-surgical samples, mainlyaffecting spayed, middle-aged and older female cats. Spaying at anearly age, has a sparing effect and reduces the risk of developingmammary tumors, but the degree of protection in cats is less pre-cisely documented than that for dogs [55]. Histologically, most fe-line mammary adenocarcinomas entered in our study weretubular and/or papillary (CTP+CT+CP, 13/21 cases) being repre-sentative of the most frequent clinical-pathological features ofFMC.

We demonstrate the presence of CSC in FMC surgical speci-mens and the possibility to isolate and culture these cells inserum-free medium, as expected for CSC.

In analogy with studies on human breast CSCs [13,21,22], weused a culture system in which FMC cells survived and proliferat-ed as non-adherent mammospheres enriched in progenitor cells,expressing CD44, but retained the differentiative potential in thepresence of serum. We also show the formation of secondarymammospheres, a defining properties of stem cells that is consid-ered as a proof of self-renewal [56][57]. Some authors reportedhuman breast cancer sphere formation from a single cell in low-attachment plates [23]. In our experimental model we were un-able to observe such an extreme result, but we obtained spheroidsfrom almost all the cultures analyzed, while in few FMC cultureswe did not detect spherogenesis but adherent packed colonies. It

100μμm50μm

A

B

C

28 30 32 34 36 38 40 42 44 46 48 50 52 54 56

50

100

150

200

250

300

Tu

mo

r V

olu

me

(% v

s. d

ay 3

0)

Time (days after injection)

1cm

Par

enta

l tu

mo

rX

eno

gra

ft t

um

or

Sphere-derived cells

Differentiated cells

Fig. 5 – Tumor formation in NOD/SCID mice injected with FMCcells. A. Phase-contrast microscopy of spheroid (left) andserum-cultured (right) cells. B. Growth curves of xenografttumors in mice. Sphere-derived cells from 2 different FMCs(red and black lines) give rise to palpable tumors (>3 mm)after 30 days following s.c. injection that significantly grow upto day 54, whereas tumor did not develop in mice implantedwith the same number of corresponding differentiated cells(green and blue lines). Dotted line is set at 100% (tumorvolume on day 30). Points represent the mean of tumors fromboth flanks of NOD-SCID mice, *p<0.05.The insect depicts arepresentative image of the tumor excised from mouse at theend of the experiment.C. H&E staining of the xenograft tumorthat closely resembles the parental feline primary tumor inmorphology and histological features (magnification 20X and40X).Negative controls (as depicted in Fig. 1) were performedas described in Material & Methods (data not shown).

856 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

is worth to note that we did not use ultra-low attachment surfaceplates, as frequently described in literature, thus our observationreflects the true sphere-forming ability of FMC cells cultured instem-permissive medium. Moreover, after detachment and re-plating, both mammospheres and compact colonies retainedself-renewal ability for multiple passages. It can be hypothesizedthat efficiency in mammosphere formation and their morphologydepend on cell type-heterogeneity and individual properties ofstem cells in each FMC sample [58]. Although mammosphere for-mation may serve as in vitro assay to identify stem cells, it was notdemonstrated in several breast cancer cell line in previous studies[29,59], supporting the relevance of the experimental model wereport in this study.

Expression of the cell-surface hyaluronan receptor CD44, andno or low expression of the membrane protein CD24, definehuman breast CSCs [13,21,60]. Most cells in FMC mammospheresexpress CD44, although negative cells were also detected. This ob-servation confirms that this cell fraction represents a subpopula-tion within CSC as previously reported [13]. CD44 expressiondecreased when FMC stem cells are differentiated in serum-containing medium, and EGF/bFGF withdrawal. Importantly, in asignificant proportion of FMC histological sections we identifiedsmall areas of CD44 expressing cells, indicating that these cellsare present in the original tumors and then are expanded invitro. Moreover, the localization of these cells in scattered areasand not spread within all the tumor, likely caused an under esti-mation of the number of expressing tumors.

Differentiation ability is also a hallmark of stemness. Cell derivedfrom freshly isolated FMCs, plated in EGF and bFGF-containing me-dium, differentiate upon growth factor withdrawal and the additionof serum. Differentiated CSCs grow as adherent monolayer, acquireepithelial-like morphology and overexpress the epithelial markerEMA, Conversely, the expression of CK18, likely indicating thepresence of luminal progenitors [48], was similar in both stem anddifferentiated FMC cells. These features, associated to a slower pro-liferation rate than CSC, indicates a near terminal differentiation inserum-containing medium. To rule out the possibility that such alimited in vitro growth resulted fromunfavorable culture conditions,we compared the in vivo tumorigenic potential of putative stem anddifferentiated FMC cells: the xenotransplant of CSC cells inducedtumor development, while differentiated cells were unable to sus-tain tumor growth in NOD-SCID mice. The tumor-initiating abilityof mammosphere definitely demonstrate the stem nature ofthese FMC cultures, being the tumorigenic ability the main distinc-tive feature of TICs, and xenograft assay the only currently reliableexperimental proof.

This is the first report showing the TIC activity of a FMC cellsubpopulation. In fact, only one study described the in vivo en-graftment of putative stem cells derived from canine mammarycarcinomas, but no comparison with differentiated cells wasreported [36]. In addition, also in human breast cancer studies,few papers report engrafting of cancer stem cell from tumor sam-ples, while the majority used sorted cells from breast cancer celllines.

We also immunohistochemically characterized EGFR and ER-alpha expression in FMCs. EGFR was associated with tumorigene-sis and representing a prognostic factor for human breast cancer.EGFR expression has been evaluated in canine [61] and, less exten-sively, in FMC cells [62] although a potential role in classification,pathogenesis and therapy warrants further investigation. EGFR

expression has been localized both in membrane/cytoplasm andnucleus in different cancer types [63,64]. Indeed, emerging evi-dence suggest the existence of a EGFR-signaling pathway inwhich activated EGFR undergoes nuclear translocation and subse-quently triggers gene expression [65] and the prognostic value ofthis mechanism has been demonstrated in human breast cancer[66]. However, no studies on animal mammary cancers analyzedEGFR localization. In our series of FMCs, EGFR expression wasdetected in 53% cases, while nuclear EGFR was detected in the14/16 tumors with 8 tissues having nuclear EGFR staining withoutcytoplasmic positivity. The discrepancy between the number ofcases expressing cytoplasmic and nuclear EGFR suggests that intumors with low or absent cytoplasmic expression, the mecha-nisms involved in receptor translocation into the nucleus mightbe more important in signal transduction, possibly leading totumor proliferation and more aggressive behavior. These data, al-though preliminary, may be relevant for further investigations.

ER-α expression is routinely evaluated in human breast tu-mors, as prognostic and therapeutic parameter [67] being ER-negative tumors more aggressive and resistant to anti-hormonaltherapy. Given the biologic similarities of human and feline mam-mary carcinomas, several studies investigated ER-α in spontane-ous feline mammary lesions [8,68] reporting that ER-αexpression dramatically decreased in high-grade tumors, althoughalmost all tumors are ER-negative. Our study, involving mainlygrade 3 tumors, showed a predominance of ER-negative (86%) le-sions, confirming the aggressive nature of these neoplasms in cats.The lack of ER-α expression may reflect the undifferentiated stateand, hence, more aggressive biologic behavior of FMC, further sup-porting FMC as model to study human ER-negative tumors.

We demonstrated EGFR and ER-α expression in FMC stem cul-tures and mammospheres. We observed a marked positivity forEGFR, supporting the relevance of this receptor expression andsignaling pathway in the survival of TICs [45]. The expression ofER-α in FMC stem-like cells although surprisingly for cells mainlyderived from ER-negative tumors could be due to a ER-positivecell expansion modulated by culture conditions as observed byFillmore [27]. However, ER-alpha expression in CSCs is still contro-versy since mammospheres from in situ ductal carcinomas do ex-press ER [69], while a low expression was reported in CD44positive cells [70]. In addition, the human breast stem cell popula-tion is enriched for steroid receptor-positive cells [71], suggestingthat the potential stem cell pool includes ER-α positive cells. Im-portantly the inhibition of this receptor may reduce the self-renewal potential of human breast CSCs [72]. Thus, although a for-mal demonstration was not provided in this study, we proposethat while the bulk of FMC cells are ER-α negative, TICs may retainthe expression of this receptor, as demonstrated by the highernumber of positive cells observed in vitro in TIC-enriched cultures

One of the main points raised by the present study is the obser-vation that the same protocol designed for the human/rodent sys-tems allows for the isolation and enrichment of cells with stemproperties from companion animals, confirming the promisingtranslational potential of the comparative oncology approach.

Conclusions

Our study demonstrates that cancer cells derived from FMC harbora population of stem-like cells that can be cultured in a defined

857E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

medium being able of self-renewing. These cells have the ability todifferentiate and to initiate tumor in vivo. Considering the repro-ducible isolation and expansion of this stem-like cell population,it may represent an attractive source for mammary CSCs for re-search and a model system for pre-clinical pharmacologicalstudies. These data suggest that the application of CSC isolationtechniques to feline tumor models, may provide an experimentaltool to develop new insights into the nature of the different celltypes in the mammary gland, the development of mammary tu-mors and the role of TICs. Spontaneous feline cancer models, al-though valuable are still under-used resource, and isolation ofTICs that can be propagated in vitro, may be useful both for basicand translational research.

Competing interestsThe authors declare that they have no competing interests.

Acknowledgments

This work was supported by grants from CARIGE Foundation(2010) and Ricerca corrente IZS-PLV (2007-2008-2009)

R E F E R E N C E S

[1] M. Paoloni, C. Khanna, Translation of new cancer treatments frompet dogs to humans, Nat. Rev. Cancer 8 (2008) 147–156.

[2] A. Jemal, T. Murray, A. Samuels, A. Ghafoor, E. Ward, M.J. Thun,Cancer statistics, 2003, CA Cancer J. Clin. 53 (2003) 5–26.

[3] A.A. Hayes, S. Mooney, Feline mammary tumors, Vet. Clin. NorthAm. 15 (1985) 513–520.

[4] E.G. MacEwen, Spontaneous tumors in dogs and cats: models forthe study of cancer biology and treatment, Cancer Metastasis Rev.9 (1990) 125–136.

[5] K. Hansen, C. Khanna, Spontaneous and genetically engineeredanimal models; use in preclinical cancer drug development,Eur. J. Cancer 40 (2004) 858–880.

[6] D.M. Vail, E.G. MacEwen, Spontaneously occurring tumors ofcompanion animals as models for human cancer, Cancer Invest.18 (2000) 781–792.

[7] J.M. de las Mulas, M. van Niel, Y. Millan, M.A. Blankenstein, F. vanMil, W. Misdorp, Immunohistochemical analysis of estrogenreceptors in feline mammary gland benign and malignantlesions: comparison with biochemical assay, Domest. Anim.Endocrinol. 18 (2000) 111–125.

[8] G.P. Burrai, S.I. Mohammed, M.A. Miller, V. Marras, S. Pirino, M.F.Addis, S. Uzzau, E. Antuofermo, Spontaneous feline mammaryintraepithelial lesions as a model for human estrogenreceptor- and progesterone receptor-negative breast lesions,BMC Cancer 10 (2010) 156.

[9] Y. Murakami, S. Tateyama, A. Rungsipipat, K. Uchida, R.Yamaguchi, Immunohistochemical analysis of cyclin A, cyclin D1and P53 in mammary tumors, squamous cell carcinomas andbasal cell tumors of dogs and cats, J. Vet. Med. Sci. 62 (2000)743–750.

[10] R. De Maria, M. Olivero, S. Iussich, M. Nakaichi, T. Murata, B.Biolatti, M.F. Di Renzo, Spontaneous feline mammary carcinomais a model of HER2 overexpressing poor prognosis human breastcancer, Cancer Res. 65 (2005) 907–912.

[11] G.T. Bergkvist, D.A. Yool, Epidermal growth factor receptor as atherapeutic target in veterinary oncology, Vet. Comp. Oncol. 9(2010) 81–94.

[12] P.C. Hermann, S. Bhaskar, M. Cioffi, C. Heeschen, Cancer stem cellsin solid tumors, Semin. Cancer Biol. 20 (2011) 77–84.

[13] M. Al-Hajj, M.S. Wicha, A. Benito-Hernandez, S.J. Morrison, M.F.Clarke, Prospective identification of tumorigenic breast cancercells, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 3983–3988.

[14] A.T. Collins, P.A. Berry, C. Hyde, M.J. Stower, N.J. Maitland,Prospective identification of tumorigenic prostate cancer stemcells, Cancer Res. 65 (2005) 10946–10951.

[15] C.A. O'Brien, A. Pollett, S. Gallinger, J.E. Dick, A human coloncancer cell capable of initiating tumour growth inimmunodeficient mice, Nature 445 (2007) 106–110.

[16] C. Li, D.G. Heidt, P. Dalerba, C.F. Burant, L. Zhang, V. Adsay, M.Wicha, M.F. Clarke, D.M. Simeone, Identification of pancreaticcancer stem cells, Cancer Res. 67 (2007) 1030–1037.

[17] S. Singh, B.S. Dwarakanath, T.L. Mathew, DNA ligandHoechst-33342 enhances UV induced cytotoxicity in humanglioma cell lines, J. Photochem. Photobiol. 77 (2004) 45–54.

[18] E. Quintana, M. Shackleton, M.S. Sabel, D.R. Fullen, T.M. Johnson,S.J. Morrison, Efficient tumour formation by single humanmelanoma cells, Nature 456 (2008) 593–598.

[19] T. Florio, Adult pituitary stem cells: from pituitary plasticity toadenoma development, Neuroendocrinology 94 (2011) 265–277.

[20] J. Stingl, Detection and analysis of mammary gland stem cells,J. Pathol. 217 (2009) 229–241.

[21] D. Ponti, A. Costa, N. Zaffaroni, G. Pratesi, G. Petrangolini, D.Coradini, S. Pilotti, M.A. Pierotti, M.G. Daidone, Isolation and invitro propagation of tumorigenic breast cancer cells with stem/progenitor cell properties, Cancer Res. 65 (2005) 5506–5511.

[22] R.B. Clarke, E. Anderson, A. Howell, C.S. Potten, Regulation ofhuman breast epithelial stem cells, Cell Prolif. 36 (Suppl 1)(2003) 45–58.

[23] G. Dontu, M. Al-Hajj, W.M. Abdallah, M.F. Clarke, M.S. Wicha,Stem cells in normal breast development and breast cancer, CellProlif. 36 (Suppl 1) (2003) 59–72.

[24] M.J. Liao, C.C. Zhang, B. Zhou, D.B. Zimonjic, S.A. Mani, M. Kaba, A.Gifford, F. Reinhardt, N.C. Popescu, W. Guo, E.N. Eaton, H.F.Lodish, R.A. Weinberg, Enrichment of a population of mammarygland cells that form mammospheres and have in vivorepopulating activity, Cancer Res. 67 (2007) 8131–8138.

[25] T.M. Phillips, W.H. McBride, F. Pajonk, The response ofCD24(−/low)/CD44+ breast cancer-initiating cells to radiation,J. Natl. Cancer Inst. 98 (2006) 1777–1785.

[26] X. Li, V. Ling, P.C. Li, Same-single-cell analysis for the study ofdrug efflux modulation of multidrug resistant cells using amicrofluidic chip, Anal. Chem. 80 (2008) 4095–4102.

[27] C.M. Fillmore, C. Kuperwasser, Human breast cancer cell linescontain stem-like cells that self-renew, give rise tophenotypically diverse progeny and survive chemotherapy,Breast Cancer Res. 10 (2008) R25.

[28] M. Zoller, CD44: can a cancer-initiating cell profit from anabundantly expressed molecule? Nat. Rev. 11 (2011) 254–267.

[29] M.J. Grimshaw, L. Cooper, K. Papazisis, J.A. Coleman, H.R.Bohnenkamp, L. Chiapero-Stanke, J. Taylor-Papadimitriou, J.M.Burchell, Mammosphere culture of metastatic breast cancer cellsenriches for tumorigenic breast cancer cells, Breast Cancer Res.10 (2008) R52.

[30] L.Y. Pang, D.J. Argyle, Using naturally occurring tumours in dogsand cats to study telomerase and cancer stem cell biology,Biochim. Biophys. Acta 1792 (2009) 380–391.

[31] B. Cogliati, T.P. Aloia, R.V. Bosch, V.A. Alves, F.J.Hernandez-Blazquez, M.L. Dagli, Identification of hepaticstem/progenitor cells in canine hepatocellular and cholangiocellularcarcinoma, Vet. Comp. Oncol. 8 (2010) 112–121.

[32] P. Matsuzaki, B. Cogliati, D.S. Sanches, L.M. Chaible, K.C. Kimura,T.C. Silva, M.A. Real-Lima, F.J. Hernandez-Blazquez, R.Laufer-Amorim, M.L. Dagli, Immunohistochemical characterizationof canine prostatic intraepithelial neoplasia, J. Comp. Pathol. 142(2009) 84–88.

[33] G. Stoica, G. Lungu, H. Martini-Stoica, S. Waghela, J. Levine, R.Smith III, Identification of cancer stem cells in dog glioblastoma,Vet. Pathol. 46 (2009) 391–406.

858 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

[34] Y. Nemoto, T. Maruo, T. Sato, T. Deguchi, T. Ito, H. Sugiyama, T.Ishikawa, H. Madarame, T. Watanabe, T. Shida, H. Sahara,Identification of cancer stem cells derived from a canine lungadenocarcinoma cell line, Vet. Pathol. 48 (2011) 1029–1034.

[35] H. Wilson, M. Huelsmeyer, R. Chun, K.M. Young, K. Friedrichs, D.J.Argyle, Isolation and characterisation of cancer stem cells fromcanine osteosarcoma, Vet. J. 175 (2008) 69–75.

[36] C. Cocola, P. Anastasi, S. Astigiano, E. Piscitelli, P. Pelucchi, L.Vilardo, G. Bertoli, M. Beccaglia, M.C. Veronesi, S. Sanzone, O.Barbieri, R.A. Reinbold, G.C. Luvoni, I. Zucchi, Isolation ofcanine mammary cells with stem cell properties andtumour-initiating potential, Reprod. Domest. Anim. 44 (Suppl 2)(2009) 214–217.

[37] M. Michishita, R. Akiyoshi, H. Yoshimura, T. Katsumoto, H.Ichikawa, K. Ohkusu-Tsukada, T. Nakagawa, N. Sasaki, K.Takahashi, Characterization of spheres derived from caninemammary gland adenocarcinoma cell lines, Res. Vet. Sci. 91(2011) 254–260.

[38] C. Penzo, M. Ross, R. Muirhead, R. Else, D.J. Argyle, Effect ofrecombinant feline interferon-omega alone and in combinationwith chemotherapeutic agents on putative tumour-initiatingcells and daughter cells derived from canine and feline mammarytumours, Vet. Comp. Oncol. 7 (2009) 222–229.

[39] J.L. Rowell, D.O. McCarthy, C.E. Alvarez, Dog models of naturallyoccurring cancer, Trends Mol. Med. 17 (2011) 380–388.

[40] A. Porrello, P. Cardelli, E.P. Spugnini, Oncology of companionanimals as a model for humans. an overview of tumor histotypes,J. Exp. Clin. Cancer Res. 25 (2006) 97–105.

[41] N.P. Nguyen, F.S. Almeida, A. Chi, L.M. Nguyen, D. Cohen, U. Karlsson,V. Vinh-Hung, Molecular biology of breast cancer stem cells:potential clinical applications, Cancer Treat. Rev. 36 (2010) 485–491.

[42] W. Misdorp, A. Romijn, A.A. Hart, Feline mammary tumors: acase–control study of hormonal factors, Anticancer Res. 11(1991) 1793–1797.

[43] D.J. Meuten, M.B. Calderwood Mays, R.C. Dillman, B.J. Cooper, B.A.Valentine, F.P. Kuhajda, D.A. Pass, Canine laryngealrhabdomyoma, Vet. Pathol. 22 (1985) 533–539.

[44] T. Florio, F. Barbieri, R. Spaziante, G. Zona, L.J. Hofland, P.M. vanKoetsveld, R.A. Feelders, G.K. Stalla, M. Theodoropoulou, M.D.Culler, J. Dong, J.E. Taylor, J.P. Moreau, A. Saveanu, G. Gunz, H.Dufour, P. Jaquet, Efficacy of a dopamine-somatostatin chimericmolecule, BIM-23A760, in the control of cell growth fromprimary cultures of human non-functioning pituitary adenomas:a multi-center study, Endocr. Relat. Cancer 15 (2008) 583–596.

[45] F. Griffero, A. Daga, D. Marubbi, M.C. Capra, A. Melotti, A.Pattarozzi, M. Gatti, A. Bajetto, C. Porcile, F. Barbieri, R.E. Favoni,M. Lo Casto, G. Zona, R. Spaziante, T. Florio, G. Corte, Differentresponse of human glioma tumor-initiating cells to epidermalgrowth factor receptor kinase inhibitors, J. Biol. Chem. 284(2009) 7138–7148.

[46] F. Barbieri, A. Bajetto, R. Stumm, A. Pattarozzi, C. Porcile, G. Zona,A. Dorcaratto, J.L. Ravetti, F. Minuto, R. Spaziante, G. Schettini, D.Ferone, T. Florio, Overexpression of stromal cell-derived factor 1and its receptor CXCR4 induces autocrine/paracrine cellproliferation in human pituitary adenomas, Clin. Cancer Res. 14(2008) 5022–5032.

[47] F. Barbieri, A. Pattarozzi, M. Gatti, C. Aiello, A. Quintero, G.Lunardi, A. Bajetto, A. Ferrari, M.D. Culler, T. Florio, DifferentialEfficacy of Sstr1, 2 and 5 Agonists in the Inhibition of C6 GliomaGrowth in Nude Mice, Am. J. Physiol. Endocrinol. Metab. 297(2009) E1078–E1088.

[48] S.Y. Park, M. Gonen, H.J. Kim, F. Michor, K. Polyak, Cellular andgenetic diversity in the progression of in situ human breastcarcinomas to an invasive phenotype, J. Clin. Invest. 120 (2010)636–644.

[49] M.H. Tomasson, Cancer stem cells: a guide for skeptics, J. Cell.Biochem. 106 (2009) 745–749.

[50] H. Clevers, The cancer stem cell: premises, promises andchallenges, Nat. Med. 17 (2011) 313–319.

[51] P.J. Keller, A.F. Lin, L.M. Arendt, I. Klebba, A.D. Jones, J.A. Rudnick,T.A. Dimeo, H. Gilmore, D.M. Jefferson, R.A. Graham, S.P. Naber, S.Schnitt, C. Kuperwasser, Mapping the cellular and molecularheterogeneity of normal and malignant breast tissues andcultured cell lines, Breast Cancer Res. 12 (2010) R87.

[52] E. Charafe-Jauffret, C. Ginestier, F. Iovino, J. Wicinski, N. Cervera,P. Finetti, M.H. Hur, M.E. Diebel, F. Monville, J. Dutcher, M. Brown,P. Viens, L. Xerri, F. Bertucci, G. Stassi, G. Dontu, D. Birnbaum, M.S.Wicha, Breast cancer cell lines contain functional cancer stemcells with metastatic capacity and a distinct molecular signature,Cancer Res. 69 (2009) 1302–1313.

[53] L. Munson, A. Moresco, Comparative pathology of mammarygland cancers in domestic and wild animals, Breast Dis. 28 (2007)7–21.

[54] T.M. Blacking, M. Waterfall, D.J. Argyle, CD44 is associated withproliferation, rather than a specific cancer stem cell population,in cultured canine cancer cells, Vet. Immunol. Immunopathol.141 (2011) 46–57.

[55] K.U. Sorenmo, R. Rasotto, V. Zappulli, M.H. Goldschmidt,Development, anatomy, histology, lymphatic drainage, clinicalfeatures, and cell differentiation markers of canine mammarygland neoplasms, Vet. Pathol. 48 (2011) 85–97.

[56] D. Dey, M. Saxena, A.N. Paranjape, V. Krishnan, R. Giraddi, M.V.Kumar, G. Mukherjee, A. Rangarajan, Phenotypic and functionalcharacterization of human mammary stem/progenitor cells inlong term culture, PLoS One 4 (2009) e5329.

[57] J.C. Lawson, G.L. Blatch, A.L. Edkins, Cancer stem cells in breastcancer and metastasis, Breast Cancer Res. Treat. 118 (2009)241–254.

[58] X. Zhao, G.K. Malhotra, S.M. Lele, M.S. Lele, W.W. West, J.D. Eudy,H. Band, V. Band, Telomerase-immortalized human mammarystem/progenitor cells with ability to self-renew and differentiate,Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 14146–14151.

[59] A. Dittmer, D. Schunke, J. Dittmer, PTHrP promotes homotypicaggregation of breast cancer cells in three-dimensional cultures,Cancer Lett. 260 (2008) 56–61.

[60] G. Rappa, A. Lorico, Phenotypic characterization ofmammosphere-forming cells from the human MA-11 breastcarcinoma cell line, Exp. Cell Res. 316 (2010) 1576–1586.

[61] A. Gama, F. Gartner, A. Alves, F. Schmitt, Immunohistochemicalexpression of Epidermal Growth Factor Receptor (EGFR) incanine mammary tissues, Res. Vet. Sci. 87 (2009) 432–437.

[62] J.M. Minke, E. Schuuring, R. van den Berghe, J.A. Stolwijk, J.Boonstra, C. Cornelisse, J. Hilkens, W. Misdorp, Isolation of twodistinct epithelial cell lines from a single feline mammarycarcinoma with different tumorigenic potential in nude mice andexpressing different levels of epidermal growth factor receptors,Cancer Res. 51 (1991) 4028–4037.

[63] C. Li, M. Iida, E.F. Dunn, A.J. Ghia, D.L. Wheeler, Nuclear EGFRcontributes to acquired resistance to cetuximab, Oncogene 28(2009) 3801–3813.

[64] H.S. Wiley, Trafficking of the ErbB receptors and its influence onsignaling, Exp. Cell Res. 284 (2003) 78–88.

[65] H.W. Lo, S.C. Hsu, M.C. Hung, EGFR signaling pathway inbreast cancers: from traditional signal transduction to directnuclear translocalization, Breast Cancer Res. Treat. 95 (2006)211–218.

[66] H.W. Lo, W. Xia, Y. Wei, M. Ali-Seyed, S.F. Huang, M.C. Hung,Novel prognostic value of nuclear epidermal growth factorreceptor in breast cancer, Cancer Res. 65 (2005) 338–348.

[67] N. Normanno, A. De Luca, P. Carotenuto, L. Lamura, I. Oliva, A.D'Alessio, Prognostic applications of gene expression signaturesin breast cancer, Oncology 77 (Suppl 1) (2009) 2–8.

[68] F. Millanta, M. Calandrella, I. Vannozzi, A. Poli, Steroid hormonereceptors in normal, dysplastic and neoplastic feline mammarytissues and their prognostic significance, Vet. Rec. 158 (2006)821–824.

[69] G. Farnie, R.B. Clarke, Mammary stem cells and breast cancer–roleof Notch signalling, Stem Cell Rev. 3 (2007) 169–175.

859E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0

Author's personal copy

[70] M. Shipitsin, L.L. Campbell, P. Argani, S. Weremowicz, N.Bloushtain-Qimron, J. Yao, T. Nikolskaya, T. Serebryiskaya, R.Beroukhim, M. Hu, M.K. Halushka, S. Sukumar, L.M. Parker,K.S. Anderson, L.N. Harris, J.E. Garber, A.L. Richardson, S.J.Schnitt, Y. Nikolsky, R.S. Gelman, K. Polyak, Moleculardefinition of breast tumor heterogeneity, Cancer Cell 11(2007) 259–273.

[71] R.B. Clarke, K. Spence, E. Anderson, A. Howell, H. Okano, C.S.Potten, A putative human breast stem cell population is enrichedfor steroid receptor-positive cells, Dev. Biol. 277 (2005) 443–456.

[72] J.W. Jung, S.B. Park, S.J. Lee, M.S. Seo, J.E. Trosko, K.S. Kang,Metformin represses self-renewal of the human breastcarcinoma stem cells via inhibition of estrogenreceptor-mediated OCT4 expression, PLoS One 6 (2011) e28068.

860 E X P E R I M E N T A L C E L L R E S E A R C H 3 1 8 ( 2 0 1 2 ) 8 4 7 – 8 6 0